REF: Delede code on pd < 1.2.2

python/xorbits/_mars/dataframe/base/transform.py

@@ -21,7 +21,7 @@
 from ...core import OutputType, recursive_tile
 from ...core.custom_log import redirect_custom_log
 from ...serialization.serializables import AnyField, BoolField, DictField, TupleField
-from ...utils import enter_current_session, pd_release_version, quiet_stdio
+from ...utils import enter_current_session, quiet_stdio
 from ..core import DATAFRAME_CHUNK_TYPE, DATAFRAME_TYPE
 from ..operands import DataFrameOperand, DataFrameOperandMixin
 from ..utils import (
@@ -33,8 +33,6 @@
     validate_axis,
 )
 
-_with_convert_dtype = pd_release_version < (1, 2, 0)
-
 
 class TransformOperand(DataFrameOperand, DataFrameOperandMixin):
     _op_type_ = opcodes.TRANSFORM
@@ -246,17 +244,10 @@ def _infer_df_func_returns(self, df, dtypes):
                     if self.call_agg:
                         infer_df = test_df.agg(self._func, args=self.args, **self.kwds)
                     else:
-                        if not _with_convert_dtype:
-                            infer_df = test_df.transform(
-                                self._func, *self.args, **self.kwds
-                            )
-                        else:  # pragma: no cover
-                            infer_df = test_df.transform(
-                                self._func,
-                                convert_dtype=self.convert_dtype,
-                                args=self.args,
-                                **self.kwds
-                            )
+                        infer_df = test_df.transform(
+                            self._func, *self.args, **self.kwds
+                        )
+
             except:  # noqa: E722
                 infer_df = None
 

python/xorbits/_mars/dataframe/indexing/reindex.py

@@ -32,7 +32,7 @@
     StringField,
 )
 from ...tensor import tensor as astensor
-from ...utils import lazy_import, pd_release_version
+from ...utils import lazy_import
 from ..core import INDEX_TYPE
 from ..core import Index as DataFrameIndexType
 from ..initializer import Index as asindex
@@ -42,9 +42,6 @@
 
 cudf = lazy_import("cudf")
 
-# under pandas<1.1, SparseArray ignores zeros on creation
-_pd_sparse_miss_zero = pd_release_version[:2] < (1, 1)
-
 
 class DataFrameReindex(DataFrameOperand, DataFrameOperandMixin):
     _op_type_ = opcodes.REINDEX
@@ -274,23 +271,14 @@ def _sparse_reindex(cls, inp, index=None, columns=None):
                         )
                         # convert to SparseDtype(xxx, np.nan)
                         # to ensure 0 in sparse_array not converted to np.nan
-                        if not _pd_sparse_miss_zero:
-                            sparse_array = pd.arrays.SparseArray.from_spmatrix(spmatrix)
-                            sparse_array = pd.arrays.SparseArray(
-                                sparse_array.sp_values,
-                                sparse_index=sparse_array.sp_index,
-                                fill_value=np.nan,
-                                dtype=pd.SparseDtype(sparse_array.dtype, np.nan),
-                            )
-                        else:
-                            from pandas._libs.sparse import IntIndex
-
-                            sparse_array = pd.arrays.SparseArray(
-                                data,
-                                sparse_index=IntIndex(index_shape, ind),
-                                fill_value=np.nan,
-                                dtype=pd.SparseDtype(data.dtype, np.nan),
-                            )
+                        sparse_array = pd.arrays.SparseArray.from_spmatrix(spmatrix)
+                        sparse_array = pd.arrays.SparseArray(
+                            sparse_array.sp_values,
+                            sparse_index=sparse_array.sp_index,
+                            fill_value=np.nan,
+                            dtype=pd.SparseDtype(sparse_array.dtype, np.nan),
+                        )
+
                         series = pd.Series(sparse_array, index=index)
 
                         i_to_columns[i] = series

python/xorbits/_mars/dataframe/reduction/tests/test_reduction_execution.py

@@ -311,9 +311,6 @@ def compute(data, **kwargs):
 
     # behavior of 'skew', 'kurt' differs for cases with and without level
     skip_funcs = ("skew", "kurt")
-    if pd_release_version <= (1, 2, 0):
-        # fails under pandas 1.2. see pandas-dev/pandas#38774 for more details
-        skip_funcs += ("sem",)
 
     if func_name not in skip_funcs:
         data_dict = dict((str(i), rs.rand(100)) for i in range(10))

python/xorbits/_mars/dataframe/window/rolling/aggregation.py

@@ -30,13 +30,12 @@
     StringField,
     TupleField,
 )
-from ....utils import calc_nsplits, has_unknown_shape, lazy_import, pd_release_version
+from ....utils import calc_nsplits, has_unknown_shape, lazy_import
 from ...core import DATAFRAME_TYPE
 from ...operands import DataFrameOperand, DataFrameOperandMixin
 from ...utils import build_empty_df, build_empty_series, parse_index
 
 cudf = lazy_import("cudf")
-_with_pandas_issue_38908 = pd_release_version == (1, 2, 0)
 
 
 class DataFrameRollingAgg(DataFrameOperand, DataFrameOperandMixin):
@@ -485,17 +484,6 @@ def execute(cls, ctx, op: "DataFrameRollingAgg"):
         else:
             data = inp
 
-            # fix for pandas 1.2.0
-            # see: https://github.com/pandas-dev/pandas/issues/38908
-            # df.rolling().aggregate('skew') modified original data
-            # so we copy it first for skew only
-            if (
-                _with_pandas_issue_38908
-                and op.func in ["skew", "kurt"]
-                and op.outputs[0].index[0] == 0
-            ):
-                data = data.copy()
-
         r = data.rolling(
             window=window,
             min_periods=op.min_periods,

python/xorbits/_mars/lib/groupby_wrapper.py

@@ -26,7 +26,6 @@
 
 cudf = lazy_import("cudf")
 
-_HAS_SQUEEZE = pd_release_version < (1, 1, 0)
 _HAS_DROPNA = pd_release_version >= (1, 1, 0)
 _GROUP_KEYS_NO_DEFAULT = pd_release_version >= (1, 5, 0)
 
@@ -47,7 +46,6 @@ def __init__(
         as_index=True,
         sort=True,
         group_keys=_default_group_keys,
-        squeeze=False,
         observed=False,
         dropna=True,
         grouper_cache=None,
@@ -79,7 +77,6 @@ def _is_frame_groupby(data: Any) -> bool:
         self.as_index = fill_value(as_index, "as_index", "_as_index")
         self.sort = fill_value(sort, "sort", "_sort")
         self.group_keys = fill_value(group_keys, "group_keys", "_group_keys")
-        self.squeeze = fill_value(squeeze, "squeeze")
         self.observed = fill_value(observed, "observed")
         self.dropna = fill_value(dropna, "dropna", "_dropna")
 
@@ -92,12 +89,9 @@ def _is_frame_groupby(data: Any) -> bool:
                 exclusions=exclusions,
                 as_index=as_index,
                 group_keys=group_keys,
-                squeeze=squeeze,
                 observed=observed,
                 dropna=dropna,
             )
-            if not _HAS_SQUEEZE:  # pragma: no branch
-                groupby_kw.pop("squeeze")
             if not _HAS_DROPNA:  # pragma: no branch
                 groupby_kw.pop("dropna")
 
@@ -127,7 +121,6 @@ def __getitem__(self, item):
             as_index=self.as_index,
             sort=self.sort,
             group_keys=self.group_keys,
-            squeeze=self.squeeze,
             observed=self.observed,
             dropna=self.dropna,
         )
@@ -219,7 +212,6 @@ def to_tuple(self, truncate=False, pickle_function=False):
             self.as_index,
             self.sort,
             self.group_keys,
-            self.squeeze,
             self.observed,
             self.dropna,
             getattr(getattr(self.groupby_obj, "grouper", None), "_cache", dict()),
@@ -237,7 +229,6 @@ def from_tuple(cls, tp):
             as_index,
             sort,
             group_keys,
-            squeeze,
             observed,
             dropna,
             grouper_cache,
@@ -256,7 +247,6 @@ def from_tuple(cls, tp):
             as_index=as_index,
             sort=sort,
             group_keys=group_keys,
-            squeeze=squeeze,
             observed=observed,
             dropna=dropna,
             grouper_cache=grouper_cache,
@@ -271,7 +261,6 @@ def wrapped_groupby(
     as_index=True,
     sort=True,
     group_keys=_default_group_keys,
-    squeeze=False,
     observed=False,
     dropna=True,
 ):
@@ -282,12 +271,9 @@ def wrapped_groupby(
         as_index=as_index,
         sort=sort,
         group_keys=group_keys,
-        squeeze=squeeze,
         observed=observed,
         dropna=dropna,
     )
-    if not _HAS_SQUEEZE:  # pragma: no branch
-        groupby_kw.pop("squeeze")
     if not _HAS_DROPNA:  # pragma: no branch
         groupby_kw.pop("dropna")
     # cudf currently not support observed,

python/xorbits/_mars/utils.py

@@ -761,7 +761,6 @@ def merge_chunks(chunk_results: List[Tuple[Tuple[int], Any]]) -> Any:
             as_index=v.as_index,
             sort=v.sort,
             group_keys=v.group_keys,
-            squeeze=v.squeeze,
             observed=v.observed,
         )
         return grouped.groupby_obj